Stretch-stimulated glucose uptake in skeletal muscle is mediated by reactive oxygen species and p38 MAP-kinase


Corresponding author M. A. Chambers: Department of Physiology; University of Kentucky, 800 Rose Street, Room MN-633; Lexington, KY 40536-0298, USA. Email:


Alternatives to the canonical insulin-stimulated pathway for glucose uptake are exercise- and exogenous reactive oxygen species (ROS)-stimulated glucose uptake. We proposed a model wherein mechanical loading, i.e. stretch, stimulates production of ROS to activate AMP-activated kinase (AMPK) to increase glucose uptake. Immunoblotting was used to measure protein phosphorylation; the fluorochrome probe 2′7′-dichlorofluorescin diacetate was used to measure cytosolic oxidant activity and 2-deoxy-d[1,2-3H]glucose was used to measure glucose uptake. The current studies demonstrate that stretch increases ROS, AMPKα phosphorylation and glucose transport in murine extensor digitorum longus (EDL) muscle (+121%, +164% and +184%, respectively; P < 0.05). We also demonstrate that stretch-induced glucose uptake persists in transgenic mice expressing an inactive form of the AMPKα2 catalytic subunit in skeletal muscle (+173%; P < 0.05). MnTBAP, a superoxide dismutase (SOD) mimetic, N-acteyl cysteine (NAC), a non-specific antioxidant, ebselen, a glutathione mimetic, or combined SOD plus catalase (ROS-selective scavengers) all decrease stretch-stimulated glucose uptake (P < 0.05) without changing basal uptake (P > 0.16). We also demonstrate that stretch-stimulated glucose uptake persists in the presence of the phosphatidylinositol 3-kinase (PI3-K) inhibitors wortmannin and LY294001 (P < 0.05) but is diminished by the p38-MAPK inhibitors SB203580 and A304000 (P > 0.99). These data indicate that stretch-stimulated glucose uptake in skeletal muscle is mediated by a ROS- and p38 MAPK-dependent mechanism that appears to be AMPKα2- and PI3-K-independent.

Skeletal muscle is critical for glucose homeostasis and glucose clearance. Insulin and exercise are important physiological stimulators of skeletal muscle glucose uptake. The mechanism of insulin-stimulated glucose uptake has been well characterized and is dependent on phosphatidylinositol 3-kinase (PI3-K) and its downstream target protein kinase B (Akt) (Lee et al. 1995). Exercise-stimulated glucose uptake is less understood. Studies using an in vitro preparation of isolated skeletal muscle have shown that contraction-stimulated glucose uptake is PI3K independent (Lee et al. 1995; Hayashi et al. 1998). An alternate, signalling pathway may involve reactive oxygen species (ROS) including superoxide anions, hydrogen peroxide and their redox derivatives. Low levels of exogenous ROS stimulate glucose uptake by adipocytes (Hayes & Lockwood, 1987), cardiac myocytes (Fischer et al. 1993), skeletal muscle myotubes (Fischer et al. 1993) and isolated skeletal muscles (Cartee & Holloszy, 1990; Kim et al. 2006; Higaki et al. 2008). Contractile activity leads to increased ROS production by skeletal muscle (Reid et al. 1992b; Stofan et al. 2000) and pre-treatment with a non-specific antioxidant, N-acetylcysteine (Mauvais-Jarvis et al. 2002), lowers glucose uptake in exercised mouse limb muscles (Sandstrom et al. 2006).

Mechanical loading

Mechanical stimuli, specifically contraction and stretch, increase rates of glucose uptake, free radical production and protein synthesis by muscle. There are two proposed mechanisms by which mechanical stimuli may regulate glucose uptake (Ihlemann et al. 1999; Richter et al. 2001). The first is a calcium-dependent mechanism, whereby the depolarization of the plasma and T tubule membranes preceding contraction stimulates sarcoplasmic reticulum calcium release and glucose transporter four (GLUT4) translocation. This has been described as a ‘feed-forward mechanism,’ in that glucose uptake is increased before metabolic needs develop. The second is a load-dependent mechanism whereby the strain put on the muscle or force developed by the muscle elicits a ‘feedback mechanism’ closely associated with metabolic needs. During contraction, the muscle is activated, calcium changes rapidly, ATP consumption is high, metabolic by-products accumulate, and ROS are produced. In contrast, stretch does not activate voltage-dependent calcium release and myofilament interactions are minimal, lessening the potential contribution of calcium- and metabolic-related changes. Ihlemann et al. (1999) have tested the effect of force on insulin-independent glucose uptake and reported that contraction-induced muscle glucose uptake varies directly with force development during tetanic contractions and that stretch increases glucose uptake. We therefore were interested in studying the signalling mechanism by which force directs glucose uptake distinct from calcium- and metabolic-related events that occur during contraction.

Muscle-derived ROS and glucose uptake

Skeletal muscle continually produces ROS at low levels under resting conditions (Reid et al. 1992b; Murrant et al. 1999) and at higher levels during contractile activity (Reid et al. 1992b). ROS represent a cascade of low molecular weight oxygen derivatives whose effects on cellular function are concentration dependent. ROS over-production has been suggested to cause oxidative damage to cellular function especially in many pathological conditions including diabetes (Yu, 1994) where oxidative stress has been linked to insulin resistance (Yu, 1994; Bonnefont-Rousselot, 2002). In addition to their pathological role, low levels (nanomolar to micromolar) of ROS may participate as second messengers in intracellular signal transduction pathways (Finkel, 1998; Stofan et al. 2000) including glucose transport signalling (Hayes & Lockwood, 1987; Cartee & Holloszy, 1990; Fischer et al. 1993; Kozlovsky et al. 1997; Sandstrom et al. 2006). Isolated mouse EDL muscles increased 2-deoxyglucose (2-DG) uptake during repetitive tetanic contractions (Sandstrom et al. 2006). It is well established that contraction leads to increased endogenous ROS production (Reid et al. 1992b). N-acetylcysteine (NAC), an antioxidant that opposes ROS action, inhibits contraction-induced glucose uptake without altering basal 2-DG uptake or uptake stimulated by insulin or hypoxia (Sandstrom et al. 2006). In aggregate, these observations suggest that glucose uptake may be increased by mechanically stimulated oxidant production.

AMPK as a downstream mediator

AMP-activated protein kinase (AMPK) is a proposed regulator of glucose uptake in exercising muscle. AMPK is a heterotrimeric serine/threonine kinase composed of a catalytic α subunit and two regulatory β and γ subunits (Mitchelhill et al. 1994; Hardie et al. 1998; Kemp et al. 1999). Each subunit has two or more different isoforms (Hardie et al. 1998). The AMPKα1 isoform, which is ubiquitously expressed (Stapleton et al. 1996), requires intense muscle contraction for activation (Hayashi et al. 2000; Musi et al. 2001b). Contrary to the α1 subunit, the AMPKα2 isoform is predominantly found in liver, heart and skeletal muscle (Stapleton et al. 1996) and is activated by moderate-intensity exercise (Hayashi et al. 2000; Musi et al. 2001b). We focused on the α2 subunit in our current study. Adenosine monophosphate (AMP) is a by-product of adenosine triphosphate (ATP) utilization. When ATP consumption is high and glucose levels are low, AMP levels increase. Elevated AMP binds and allosterically modifies AMPK, rendering it a better substrate for the upstream activating kinases and a less likely target for protein phosphatases. AMPK is thought to be regulated by factors that change the ratio of AMP : ATP such as hypoxia, heat shock, metabolic toxicity, exogenous ROS (Choi et al. 2001; Musi et al. 2001a; Fryer et al. 2002) and exercise (Musi et al. 2001a; Musi & Goodyear, 2003). Phosphorylation of threonine 172 (Thr172) in the activation loop of the α subunit is required for AMPK activation (Hardie et al. 1998).

p38 MAPK as a downstream mediator

p38 mitogen-activated protein kinase (MAPK) is a stress-activated protein serine/threonine kinase known to be responsive to oxidative stress (Clerk et al. 1998; Li et al. 2005). While controversial, some studies suggest p38 MAPK may mediate insulin-stimulated glucose uptake (Sweeney et al. 1999; Somwar et al. 2000). More recently, studies suggest p38 MAPK may also mediate AMPK- and/or ROS-regulated glucose uptake. Selective p38 MAPK inhibition abolished the increase in glucose transport by acute exposure to (60–90 μm) H2O2 (Kim et al. 2006). 5-Aminoimidazole-4-carboxamide ribonucleoside (AICAR)-stimulated glucose transport was inhibited by the p38 MAPK inhibitor, SB203580, and also by overexpression of a dominant-negative p38 MAPK mutant (Xi et al. 2001).

We hypothesized that mechanical loading increases muscle-derived ROS, which in turn stimulate protein kinase activity and lead to increased glucose uptake. After stretch, we measured ROS production, AMPKα, Akt and p38 MAPK phosphorylation, and glucose uptake. To determine whether these signalling events are essential for increased glucose uptake, we used pharmacological and genetic interventions that target pathway components.



2-Deoxy-d-[1,2-3H]glucose and d-[1-14C]mannitol were purchased from Perkin Elmer (Boston, MA, USA). Antibodies against pan-α-AMPK, phosphorylated α-AMPK (T172), phosphorylated Akt (Ser473) and total p38 MAPK were purchased from Cell Signalling Technologies (Danvers, MA, USA). The antibody for AMPKα2 was purchased from Abcam (Cambridge, MA, USA). Antibodies against total Akt and phosphorylated p38 MAPK (Thr180/Tyr182) were purchased from ECM Biosciences (Versailles, KY, USA). The p38 MAPK inhibitor A304000 was a gift from Abbott Laboratories (Abbott Park, IL, USA). Superoxide dismutase and catalase were from Oxis International (Foster City, CA, USA). MnTBAP and ebselen were from A.G. Scientific (San Diego, CA, USA). All other reagents were from Sigma Aldrich (St Louis, MO, USA).

Animal use

All procedures were approved by the Institutional Animal Care and Use Committee of the University of Kentucky Medical Center and were conducted in strict accordance with the Public Health Service animal welfare policy. Adult ICR mice (25–34 g; Harlan Sprague Dawley, Indianapolis, IN, USA) and transgenic mice (muscle-specific inactive α2 of AMPK; C57B/6) were housed in stainless-steel cages using a 12 h on–12 h off lighting schedule and were fasted 12–18 h overnight prior to study. Animals were anesthetized and killed by rapid cervical dislocation and the extensor digitorum longus (EDL) muscles were excised.

Muscle preparation

EDL muscles were isolated in buffered Krebs-Ringer solution (in mm: NaCl 117, KCl 4.7, CaCl2 2.5, MgSO4 1.2, NaH2PO4 1.2, NaHCO3 24.6) that contained 2 mm pyruvate and was equilibrated with 95% O2 and 5% CO2 at 37°C (pH after equilibration ∼7.35). Nylon thread was tied to both tendons and the muscles were secured between an anchored glass rod and an adjustable holder (World Precision Instruments) to which a force transducer (Kulite, Leonia, NJ, USA) was attached. A model S48 square-wave stimulator (Grass Instruments, Quincy, MA, USA) was used to deliver a 0.5 ms pulse at supramaximal voltage. Twitch force was recorded using an oscilloscope (546601B; Hewlett Packard, Palo Alto, CA, USA) and a chart recorder (BD-11E; Kipp and Zonen, Delft, the Netherlands). The length of the muscle was adjusted to maximize twitch force (optimum length Lo), and the passive force was recorded. In experiments testing the effects of stretch, muscle length was adjusted and maintained at approximately 90%Lo for the unloaded control or 120%Lo for stretch. The length of 120%Lo was selected for several reasons. We routinely measure contractile function after stretching the muscle to 120–130%Lo without compromising muscle function. When the muscle is stretched beyond 130%Lo, we start to see signs of muscle damage, such as fibres tearing. Furthermore, pilot studies indicated that oxidant activity does not change between lengths of 110% and 120%Lo; however, we see less variability at 120%Lo. For these reasons, we selected 120%Lo for our stretch model.

Cell culture

C2C12 skeletal myotube cultures were grown by plating C2C12 myoblasts on type I collagen-coated Bioflex membranes (Flexcell, Hillsborough NC, USA). C2C12 myoblasts were purchased from ATCC (Manassas, VA, USA) and were grown in Dulbecco's modified Eagle's medium supplemented with antibiotics (100 μg ml−1 streptomycin and 100 units ml−1 penicillin (Sigma)) and 10% fetal bovine serum. After 80% confluence, cells were switched to DMEM, supplemented with antibiotics and 2% heat-inactivated horse serum to promote differentiation. Cells were maintained in this medium for 48 h and then switched to the 2% fetal bovine serum medium for an additional 5–6 days, resulting in the formation of distinct multinucleated myotubes.

C2C12 myotubes grown on Bioflex membranes were subjected to 10 min of 15% intermittent (1 Hz) multi-axial stretch using a triangular-waveform (Flexercell (FX-3000) device; Flexcell) or static conditions. The 15% strain was selected based on a previous publication that showed mechanical signal transduction in response to this strain (Hornberger et al. 2005). Myotubes were collected immediately after stretch and subjected to Western blot analysis as described below.

Total protein and Western blot analysis

After treatment, myotubes were washed with PBS and scraped from the surface or muscles were homogenized by hand in 200 μl of 20 mm Tris HCl, pH 7.5, 2 mm ATP, 5 mm MgCl2 and 1 mm dithiothreitol (DTT). The lysates were sonicated on ice and then heated at 98°C for 5 min. Equal amounts of protein were loaded in each lane of 4–15% Tris-HCl polyacrylamide gels and electrophoresed at 200 V for 50 min. Proteins were either dyed using Simply Blue (Invitrogen) and scanned for total protein, or transferred at 200 mA for 2 h to nitrocellulose membranes for Western blot. Membranes were blocked in blocking buffer–PBS (Odyssey; Li-COR Biosciences, Lincoln, NE, USA) for 1 h at room temperature, incubated with primary antibodies overnight, followed by four 5 min washes. Membranes were incubated with fluorescence-conjugated secondary antibodies in Odyssey blocking buffer–PBS and 0.01% SDS for 45 min, followed by four 5 min washes. The membrane was then dried and blots were scanned by densitometer (Odyssey) to quantify differences.

Immunodepletion of AMPKα2 was performed using the Pierce Classic Immunoprecipitation Kit (Pierce Biotechnology, Rockford, IL, USA). Briefly, after muscles were incubated under basal or stretch conditions, they were immediately frozen in liquid nitrogen. The muscles were then homogenized on ice in IP lysis buffer (0.025 m Tris, 0.15 m NaCl, 0.001 EDTA, 1% NP-40, 5% glycerol, pH 7.4) and centrifuged at 1000 g for 5 min to remove cellular debris. Immune complexes were allowed to form overnight by incubating the muscle lysate (500 μg protein) with the AMPKα2 antibody (4 μg). Immune complex was captured using Pierce protein A/G agarose and eluted in reducing sample buffer; this sample buffer was also used to denature the AMPKα2-depleted fraction.

Cytosolic oxidant activity

The fluorochrome probe 2′,7′-dichlorofluorescin diacetate (DCFH-DA; Molecular Probes, Eugene, OR, USA) was used to measure oxidant activity (Reid et al. 1992a). Mature C2C12 myotubes or excised EDL muscles were loaded with DCFH-DA 20 μm, for 15–45 min. Accumulation of the oxidized derivative (DCF; 480 nm excitation, 520 nm emissions) was measured by use of an epifluorescence microscope (Labophot-2; Nikon Instruments, Melville, NY, USA), a CCD camera (Series 72, Dage-MTI, Inc., Michigan City, IN, USA), and a computer-controlled shutter in the excitation light pathway. DCF emissions were acquired by 20 ms exposure to excitation light (480 nm); records were stored to the computer for measurement of emission intensity using commercial data acquisition and analysis software (Optimas 4.02; Bioscan, Edmonds, WA, USA). Final values for DCF emission intensity were corrected for photo-oxidation artifact.

Glucose uptake

To measure 2-deoxy-d[1,2-3H]glucose uptake, paired EDL muscles were incubated at 37°C for 30 min in Krebs bicarbonate buffer (117 mm NaCl, 4.7 mm KCl, 2.5 mm CaCl2, 1.2 mm KH2PO4, 1.2 mm MgSO4 and 24.6 mm NaHCO3, pH 7.5) containing 2 mm pyruvate and equilibrated with 95% O2–5% CO2. One muscle per pair was maintained at a stretched length of 120%Lo while the contralateral muscle was maintained at a resting length of 90%Lo. After 50 min, the buffer was drained and new buffer plus 1 mm3H-glucose (1.5 μCi ml−1)–7 mm14C-mannitol (0.45 μCi ml−1) was added to the organ baths (Perkin Elmer, Boston, MA, USA) at 37°C. When drugs were administered, they were present during all incubations. After 10 min, muscles were immersed in buffer containing no sugars, blotted, cut from threads, and frozen in a 1.5 ml microcentrifuge tube at −80°C. Frozen muscles were weighed and transferred to another microcentrifuge tube for digestion in 250 μl of 1 n NaOH. Samples were heated at 80°C for ∼10 min. Samples were vortexed and spun, and 250 μl of 1 n HCl was added to neutralize the NaOH. Samples were vortexed and spun again and 350 μl was pipetted to a minivial containing 4 ml scintillation cocktail. Samples were counted overnight in a scintillation counter set up for dual-label dpm. Transport rates for each sample were determined by dpm counts for all samples including blanks.

Statistical analysis

Data are expressed as means ±s.e.m. Statistical analyses were performed using a Student's t test or one-way ANOVA. Student's t test was used for the comparison of two means while ANOVA was used for the comparison of multiple means. When ANOVA revealed significant differences, Fisher's LSD post hoc test for multiple comparisons was performed. P values < 0.05 were considered significant.


Stretched EDL develops force

To determine the force developed by stretch, we pre-incubated EDL at Lo for 30 min at 37°C and measured maximum tetanic force (300 Hz). Thereafter, we measured the forces produced by stretching unstimulated EDL to lengths of 100–130%Lo. The resulting data were used to generate a classic length–force curve (Fig. 1). Force increased slightly from 100 to 115%Lo and then rose more steeply from 120–130%Lo. For subsequent studies of EDL function, we stretched the muscles to approximately 120%Lo. This stretch applied ∼4 N cm−2 to the muscle, or ∼10% of the force produced by maximal contraction (Po= 37 ± 3 N cm−2).

Figure 1.

Length–force curve for EDL
Force was measured at EDL lengths of 100 to 130%. The force produced by a maximal contraction was equal to 37 N cm−2. Stretch at 120%Lo applied 4 N cm−2 to the muscle (N= 3 muscles per length).

Stretch induces oxidant activity

Oxidants were detected by use of the dichlorofluorescin (DCFH) oxidation assay. DCFH is loaded into the cell where it is oxidized to a fluorescent derivative, dichloroflourescein (DCF), at a rate proportional to cytosolic oxidant activity. We measured DCF fluorescence as an index of intracellular oxidant activity in basal (Fig. 2A) and stretched EDL (Fig. 2B), and basal (Fig. 2C) and stretched myotubes (Fig. 2D). We demonstrate for the first time stretch increased oxidant activity in both EDL and myotubes. To determine the nature of the oxidant activity, we employed ROS-selective scavengers. These included the combined treatment of superoxide dismutase (SOD; degrades superoxide) plus catalase (degrades H2O2), or a glutathione mimetic, ebselen. ROS-selective inhibitors abolished the 20% change in oxidant activity induced by stretching EDL (Fig. 3).

Figure 2.

Stretch-induced oxidant activity
DCF fluorescence was measured as an index of intracellular oxidant activity. A, basal EDL muscle has low levels of oxidant activity. B, stretched EDL muscle has increased levels of oxidant activity. C, basal C2C12 cells have low levels of oxidant activity. D, stretched C2C12 cells have increased levels of oxidant activity. All pictures were optimized equally.

Figure 3.

Stretch-induced ROS
The stretched-induced increase in oxidant activity was abolished when EDL was incubated in the presence of either SOD+catalase or ebselen. Paired EDL were exposed to antioxidant treatment +/− stretch. (N= 6–8 pairs per condition; *P < 0.05 vs stretch alone).

Reactive oxygen species mediate stretch-stimulated glucose uptake

Muscle-derived ROS promote contraction-simulated glucose uptake (Sandstrom et al. 2006). To determine if ROS also promote glucose uptake during passive stretch, we measured glucose uptake by EDL in the presence of ROS-selective and non-specific antioxidants. For a positive control of AMPK-mediated glucose uptake, we stimulated the muscle with AICAR, a pharmacological activator of AMPK (Musi & Goodyear, 2003; Aschenbach et al. 2004). Glucose uptake in response to passive stretch increased 84% above basal and was 60% of AICAR-stimulated glucose uptake (Fig. 4A). These data are consistent with the magnitude of stretch-stimulated glucose uptake observed by Ihlemann et al. (1999). We next measured the contribution of ROS in stretch-stimulated glucose uptake. We used four interventions: (1) N-acetylcysteine, a non-specific antioxidant that supports glutathione synthesis (Mauvais-Jarvis et al. 2002; Ferreira & Reid, 2008), (2) SOD plus catalase (superoxide and hydrogen peroxide scavengers), (3) ebselen (a glutathione peroxidase mimetic), and (4) MnTBAP, a SOD mimetic (Fig. 4B). All interventions suppressed stretch-stimulated glucose uptake (P < 0.05) without changing basal uptake (data not shown, P > 0.16).

Figure 4.

ROS mediate stretch-induced glucose uptake
A, stretch of EDL (120%Lo) increased glucose uptake 84% above basal. AICAR treatment (2 mm) increased glucose uptake 210% above basal (N= 4–8). B, four antioxidant interventions were used to suppress stretch-stimulated glucose uptake: (1) NAC (10 mm); (2) SOD–catalase (1000 U ml−1); (3) ebselen (30 μm); or (4) MnTBAP (100 μm). Paired contralateral EDL muscles were incubated in the presence of the antioxidant with or without stretch (N= 4–8 pairs per condition; *P < 0.05 vs control without stretch, #P < 0.05 vs control with stretch).

AMPKα2 does not mediate stretch-stimulated glucose uptake

AMPK is an important regulator of glucose uptake. We found that stretch increased AMPKα phosphorylation in EDL after 1 h (Fig. 5A) and in myotubes after 15 min (Fig. 5B). We then immunodepleted AMPKα2 and detected no change in phosphorylation of the remaining AMPK (98 ± 21.87% control; N= 4, P > 0.69). These findings suggest the α2 isoform is preferentially phosphorylated in response to stretch. To determine if AMPKα2 mediates stretch-stimulated glucose uptake, we studied AMPKα2i transgenic mice that express an inactive α2 catalytic subunit (Fujii et al. 2005). Stretch increased glucose uptake 71% in the wildtype EDL and 72% in the AMPKα2i EDL compared to their unstretched genetic controls (Fig. 5C). These results indicate that AMPKα2 activity is not essential for stretch-stimulated glucose uptake.

Figure 5.

Stretch and AMPK signalling
A, EDL was stretched over time, and AMPK phosphorylation (Thr 172, pAMPK) was measured and normalized for total AMPK (pan-α-AMPK, tAMPK) in control EDL muscle and EDL stretched for 5–60 min (N= 3 pairs per time). B, C2C12 cells were stretched for 10 min, and AMPK phosphorylation was measured and normalized for total AMPK (N= 6 per condition). C, stretch-induced glucose uptake was measured in EDL from littermate wildtype and AMPKα2i transgenic mice. Paired contralateral EDL muscles were incubated with and without stretch (N= 3–7; *P < 0.05 vs control).

PI3-kinase/Akt signalling and stretch-stimulated glucose uptake

PI3-K/Akt signalling regulates insulin-stimulated glucose uptake (Lee et al. 1995; Cho et al. 2001). We tested whether stretch stimulates Akt phosphorylation and whether stretch-stimulated glucose uptake is regulated by PI3-K. Stretch increased Akt phosphorylation by 23% in EDL muscle (Fig. 6A) and by 50% in C2C12 cells (Fig. 6B). To test the contribution of PI3-kinase, we measured glucose uptake in the presence of a PI3-kinase inhibitor, wortmannin or LY294002. Post hoc comparisons of data sets from different experiments (Figs 4Avs 6C) suggest the change in glucose uptake caused by stretch may be greater under drug-free conditions (+84%vs control) than in muscles pre-treated with LY294002 (+59%; P < 0.05 vs drug free). This was not true for muscles pre-treated with wortmannin (+95%; NS).

Figure 6.

Stretch and PI3-K/Akt signalling
A, EDL was stretched over time, and Akt phosphorylation (Ser 473) was measured and normalized for total Akt (tAkt, N= 3 pairs per time). B, C2C12 cells were stretched for 10 min, and Akt phosphorylation (Ser 473) was measured and normalized for total Akt, N= 6 per condition. C, stretch-stimulated glucose uptake was measured in EDL in the presence of PI3-K inhibitors, wortmannin (500 nm) or LY294002 (20 μm), for both control and stretched muscles. Comparisons are between paired control (unstretched) and stretch muscles in the presence of the inhibitor (N= 4–6; *P < 0.05 vs control + inhibitor).

p38 MAP-kinase mediates stretch-stimulated glucose uptake

Exercise and contraction stimulate p38 MAPK signalling in skeletal muscle (Ryder et al. 2000; Yu et al. 2001). We therefore tested whether p38 MAPK might regulate stretch-stimulated glucose uptake. In EDL, p38 MAPK phosphorylation increased within 5 min and remained elevated up to 1 h (Fig. 7A). This response was a faster and larger response than we observed for either Akt or AMPK phosphorylation. We also observed a 60% increase in p38 MAPK phosphorylation in C2C12 cells in response to stretch (Fig. 7B). To test the contribution of p38 MAPK to glucose uptake, we used a p38 MAPK inhibitor, SB203580 or A304000. Each inhibitor blocked the increase in glucose uptake during stretch (Fig. 7C). Thus, p38 MAPK appears to mediate stretch-induced glucose uptake.

Figure 7.

p38 MAPK signalling
A, EDL was stretched for 5–60 min, and p38 MAPK phosphorylation (pp38; Thr 180/Tyr 182) was measured and normalized for total p38 MAPK (tp38; N= 3 pairs per time). B, C2C12 cells were stretched for 10 min and p38 MAPK phosphorylation (Thr 180/Tyr 182) was measured and normalized for total p38 MAPK (N= 6 per condition). C, glucose uptake by stretched and unstretched EDL was measured in the presence of p38 MAPK inhibitors SB203580 (10 μm) or A304000 (5 μm). Comparisons are between paired control (unstretched) and stretched muscles in the presence of each inhibitor (N= 4).


The present study demonstrates that stretch increases glucose uptake in skeletal muscle. This response appears to be mediated by endogenous ROS and p38 MAPK signalling. Our data provide no support for PI3-K/Akt or AMPKα2 involvement.

Early studies have shown that force developed during contraction is linearly proportional to glucose uptake (Ihlemann et al. 1999; Fujii et al. 2005). In these studies, the investigators shortened the muscle or decreased the voltage of electrical stimulation so that the muscle generated less force, and concurrently observed decreased contraction-stimulated glucose uptake. Sandstrom et al. recently reported that mechanical loading plays little role in contraction-mediated glucose uptake (Sandstrom et al. 2007). In this study, the authors used N-benzyl-p-toluene sulphonamide (BTS, an inhibitor of myosin II ATPase) to block crossbridge activity, thereby inhibiting force production. Sandstrom et al. demonstrated that BTS reduced the force developed during contraction with no effect on glucose uptake. The difference in techniques among these three studies could explain the apparent discrepancies in their conclusion on the influence of mechanical load on glucose uptake. Ihlemann et al. altered the length of the muscle, Fujii and colleagues altered the amount of muscle activation while Sandstrom et al. inhibited cross-bridge formation to manipulate force development during contraction. In each of these studies, developed force was altered, in concert with other cellular properties including diffusion distance, cross-bridge interactions with thin filaments, metabolism, and/or calcium release. To minimize the contribution of the many cellular changes associated with contraction, we tested the contribution of mechanical load directly, via stretch. Many cellular responses differ between passive and active force development in the muscle, and perhaps it is this difference in stimuli which leads to distinct mechanical signal transduction pathways.

Skeletal muscles produce ROS at low levels under resting conditions (Reid et al. 1992b; Murrant et al. 1999) and at higher levels during contractile activity (Reid et al. 1992b). Despite the ubiquitous influence of muscle-derived ROS, few studies have tested the role of endogenous ROS on glucose uptake. Consistent with prior findings (Sandstrom et al. 2006), we demonstrate the non-specific antioxidant NAC suppresses load-stimulated glucose uptake. We further show that several ROS-specific scavengers are equally effective, suggesting ROS are the dominant oxidants responsible for stretch-stimulated glucose uptake. We demonstrate that ROS in general mediate the stretch-stimulated response. However, it is possible that NO and/or NO derivatives might also mediate the response based on published studies that already support a role for NO in insulin-independent glucose uptake (Balon & Nadler, 1997; Higaki et al. 2001).

The role of AMPK in contraction-stimulated glucose uptake has been a topic of considerable research. Early studies suggested AMPK might regulate contraction-stimulated glucose uptake: the effects of insulin and AICAR on glucose uptake are additive; the effects of AICAR and contraction are not (Hayashi et al. 1998). Further, contraction of isolated muscle activates AMPK in a load-dependent manner that parallels the contraction-associated rise in glucose uptake (Ihlemann et al. 1999; Fujii et al. 2005).

Recent studies of isoform specificity suggest AMPKα2 is not essential for contraction-stimulated glucose uptake (Mu et al. 2001; Fujii et al. 2005). Similarly, AMPKα2 does not appear to be essential for stretch-stimulated glucose uptake. Stretch clearly stimulated AMPKα phosphorylation. This response was undetectable in the AMPKα2-immunodepleted fraction, suggesting the AMPKα2 isoform is preferentially phosphorylated by stretch. However, stretch-stimulated glucose uptake was identical between wildtype and AMPKα2 inactive mice. Thus, stretch may activate AMPKα2, but this kinase does not appear to mediate the increase in glucose uptake.

Extracellular application of H2O2 appears to stimulate skeletal muscle glucose uptake through a PI3-K-dependent pathway (Kim et al. 2006; Higaki et al. 2008). Exogenous H2O2 was used to stimulate glucose uptake, which was inhibited with the PI3-K inhibitor wortmannin (Higaki et al. 2008). Kim et al. used glucose oxidase to generate H2O2 (60–90 μm) and were also able to increase glucose uptake in a PI3-K-dependent manner.

We found that muscle-derived ROS stimulate glucose uptake in response to stretch and that this increase persists in the presence of two distinct PI3-K inhibitors. Neither wortmannin nor LY-294002 abolished stretch-stimulated glucose uptake. LY-294002 did decrease the relative magnitude of the stretch response, suggesting PI3-K might partially modulate stretch-stimulated glucose uptake. However, this speculation is contradicted by the wortmannin data and is based on post hoc analyses of non-optimized experiments.

Our results appear to conflict with prior reports that extracellular H2O2 application increases glucose uptake in a PI3-K-dependent manner (Kim et al. 2006; Higaki et al. 2008). Several key differences could explain this dichotomy. First, the chemistry: muscle-derived ROS comprise a complex cascade of low molecular weight, redox-active molecules with different physical and chemical properties than H2O2. Second, the compartmentalization: exogenous H2O2 applied to the sarcolemmal surface is likely to activate different signalling cascades than ROS generated by mitochondria, NADPH oxidase or other internal sources. Third, the concentration: exogenous H2O2 was applied at concentrations a magnitude higher than the cytosolic levels predicted for internal ROS production (Higaki et al. 2008). Under our experimental conditions, PI3-K/Akt signalling does not appear to be essential for stretch-stimulated glucose uptake.

Another plausible candidate for the exercise-responsive pathway is p38 MAPK. In rat liver epithelial cells, ACIAR-stimulated glucose uptake was inhibited by the p38 MAPK inhibitor SB203580, or by overexpression of a dominant negative p38 MAPK mutant (Xi et al. 2001). In muscle, AICAR activated p38 MAPK concomitantly with AMPK, and AICAR-stimulated glucose uptake was blunted by SB203580 (Lemieux et al. 2003). A more recent study demonstrated that p38 MAPK is not a downstream component of AMPK-mediated signalling (Ho et al. 2007). These data suggest p38 MAPK is sufficient but not necessary for AICAR-stimulated glucose uptake.

In muscle, glucose uptake stimulated by exogenous ROS is dependent on p38 MAPK (Kim et al. 2006). Our results are consistent with this finding. In response to stretch, we observed early and sustained increases in p38 MAPK phosphorylation that preceded AMPK phosphorylation. Furthermore, two distinct p38 MAPK inhibitors abolished glucose uptake in response to stretch. Specificity is a question with any pharmacological inhibitor. For example, SB203580 is known to have non-specific affects on Akt phosphorylation. Therefore, we used a more specific p38 MAPK inhibitor, A304000, which inhibits p38 MAPK-mediated glucose uptake without affecting GLUT4 translocation or c-Jun N-terminal kinase (JNK) activity (Somwar et al. 2002) or altering Akt phosphorylation (Kim et al. 2006). Neither inhibitor affected basal glucose uptake. Our data suggest that p38 MAPK is necessary for stretch-dependent glucose uptake and that p38 MAPK is not a downstream component of AMPK-mediated signalling.

In summary, the current study demonstrates that the force developed during stretch is sufficient to stimulate glucose uptake and that this response is mediated by ROS. The mechanism by which stretch increases glucose uptake does not appear to involve AMPKα2 or PI3-K/Akt signalling. Instead, p38 MAPK is necessary. This novel signalling mechanism is distinct from canonical contraction- and insulin-stimulated signalling that increases glucose uptake. It may provide an alternative pathway for the development of novel therapeutic drugs to overcome insulin resistance.



We thank Dr Karyn Esser for the use of the Flexcell (FX-3000) device for the cell culture studies. This research was supported by grants from the NIH (DK066232) and (AR42238) and the American Heart Association (AHA0615263B).